242 Seed fall and flowering in white box

Seed fall and flowering in white box (Eucalyptus albens Benth.) trees near Cowra, New South Wales

W.S. Semple1,2, T.B. Koen3,4 and J. Henderson3

1Department of Environment and Climate Change, PO Box 53, Orange, NSW 2800, Australia 2Email: [email protected] 3Department of Environment and Climate Change, PO Box 445, Cowra, NSW 2794, Australia 4Corresponding author email: [email protected]

Revised manuscript received 5 September 2007

Summary has to escape predation so that it can be incorporated into a soil seed bank where it must persist until the right conditions stimulate There are few studies of the factors affecting the recruitment germination; the young plant then has to survive and grow. of white box (Eucalyptus albens), a woodland tree of the inland They noted that these processes were poorly understood in slopes of south-eastern Australia with limited recruitment in E. salmonophloia but could have added that this also applies to recent times. One of these factors, seed fall, was monitored for many other woodland eucalypts, including E. albens* (white periods ranging from 18 to 43 months (September 1995 to April box), the subject of this paper. General principles of eucalypt 1999) at 14 E. albens trees in central western New South Wales. biology, including regeneration from seed, have been built up Seed collectors were located beneath the canopies of all trees over many years in high-rainfall areas where most of the and also outside the canopies at 12 of them. Flowering was also commercial timber species of eucalypt occur. Much of this assessed — both visually and from the numbers of opercula in information has been compiled by, for example, Jacobs (1955), the collectors. Seed fall was highly variable between trees. Penfold and Willis (1961), Hillis and Brown (1984), Cremer et Although one tree shed seed most of the time, and all those that al. (1990) and Florence (1996), and with respect to seed, very were monitored for 43 months ultimately shed seed, there were comprehensively by Boland et al. (1980). As noted by Windsor lengthy periods when very little seed was collected. It appeared (2000), however, the extent to which these principles apply to that most of the early-mature capsules present on all trees in woodland species is less well studied. mid-1995 did not shed seed until 1998. Seed fall beyond the canopy was low and from the limited data collected, it appeared Observations on the inland slopes of New South Wales (NSW) that seed was evenly dispersed around the canopy. Flowering indicate that eucalypt regeneration is uncommon in grazed was similarly variable between trees, for example some paddocks compared to roadsides (Schabel and Eldridge 2001), produced abundant flowers every second year whereas others but even on the latter it is quite variable (Nowland 1997). Some flowered in two or three consecutive years. However, averaged regeneration, possibly from lignotuberous seedlings, was reported across all trees, flowering was abundant in 1997 and 1999 and (Spooner et al. 2002) following fencing of selected stands. was associated with above-average winter and spring rainfall in the previous year. Despite the presence of short-term seedbeds Factors affecting regeneration of woodland eucalypts on the during the early period of monitoring and the absence of grazing inland slopes and adjacent tablelands of south-eastern Australia at 12 of the trees, seedling recruitment was minimal during the have been described by Curtis (1989, 1990) and Windsor (2000), 3.5 y of observation. but observations on seed fall are very limited — for example Curtis (1989, 1990) and Lawrence et al. (1998). Capsule Keywords: regeneration; flowering; seed crops; seed collection; viability; dehiscence in eucalypts is initiated by twig death or the formation seed dispersal; seedbed preparation; survival; rain; Eucalytus albens of an abscission layer that cuts off the sap flow to the capsules. Fertilised ovules are shed as seed and unfertilised ones (the Introduction majority) as ‘chaff’. Seeds are dispersed by wind for short distances — up to 1.5 times the height of the tree, but this can be Eucalypt regeneration depends on processes that need to operate exceeded if trees are in exposed situations. With respect to synchronously or at particular times. It can fail if any one of the woodland eucalypts in the highly modified rural environments processes fails. In their introduction to a study of the Western of central western NSW, a number of questions arose. Is seed Australian woodland eucalypt, Eucalyptus salmonophloia fall and seed viability sufficient for seedling recruitment in most F.Muell., Yates et al. (1994, p. 532) cited J.L. Harper to explain the operation of the main processes:

For recruitment to occur, trees have to flower, the flowers have to be pollinated and set and disperse viable seed; the seed then *Except where indicated, botanical nomenclature follows that of Harden (1990–93).

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 W.S. Semple, T.B. Koen and J. Henderson 243 years? When do peaks of seed fall occur, and is recruitment Seed collectors likely to be successful at this time? Are asymmetrical distributions of seedlings around tree canopies due to uneven The design of the collectors was based on that of Curtis (1989). wind dispersal of seed as noted by Venning (1988), or are other The bottom of a plastic garbage bin was removed and a circular factors such as protection from frost or desiccation (e.g. Curtis piece of cloth mesh glued to the underside. An earlier seed- 1990) involved? collecting exercise (Lawrence et al. 1998) indicated that ‘Petlee’s polyester-cotton 2125 La Coste’ fabric was durable and retained Despite extensive clearing, E. albens is still relatively common eucalypt seed and chaff while allowing the passage of water. Mean throughout most of its former range, southern Queensland to diameter of the bin opening was 0.385 m, giving a collection northern Victoria, even though most of the area originally area of 0.116 m2. occupied by these woodlands is now cropped and/or grazed. Intact woodlands are rare (Prober 1996) but stands with relatively intact Collectors were suspended, with their tops about 1 m above the structure and moderately to highly modified groundstoreys are ground, from two steel pegs with two additional pegs serving as more common, albeit small and fragmented. The species most stabilisers (Fig. 1). A circular piece of rabbit netting was placed commonly occurs in roadside corridors and as scattered paddock on top of each collector to prevent falling twigs from damaging trees. The latter occurrence provided an opportunity to monitor the mesh and the entry of small animals. A sticky material seed fall beneath and around the canopy of individual trees in (‘Rentokil Bird Off’) was applied near the base of the droppers the absence of contamination from other trees. The specific to prevent ground-based invertebrates from gaining access to the objectives of the study were to: collector. Herbage growth beneath collectors was controlled by applications of herbicide as required. • determine the amount, viability and seasonal pattern of seed fall As cattle were expected to damage seed collectors, the 12 trees • test the hypothesis that seed is dispersed, at least over short at Walli were individually fenced so as to include the canopy and distances, evenly around the canopy an area 5 m beyond mean canopy diameter. The fenced areas • document the cycle of flowering during the period of therefore varied in size, but all were square with sides aligned observations. north-south and east-west with the trunk of the tree at the centre. As no cattle were present at the Cowra and Molong sites, the Originally, it was also intended to promote synchronous seed trees there were not fenced. Over the period of observation, fall onto a receptive seedbed by burning beneath the trees collectors sustained various types of damage, for example from (Florence 1996) in an attempt to promote seedling recruitment. falling limbs and cattle breaking down fences during dry times, This was subsequently abandoned, but relevant results are or simply wearing out. Repair or replacement was an on-going reported. activity. Missing data were therefore more common than was desired. Methods

Tree selection

The main study site was located on rolling country at Walli, about 30 km north-east of Cowra at an altitude of 560–600 m asl and with a mean annual rainfall (MAR) exceeding 650 mm. The site was largely cleared of trees, relatively fertile (enhanced by past applications of fertiliser) with a pasture (groundstorey) dominated by Phalaris aquatica and grazed by cattle. In June 1995 we searched for relatively isolated E. albens trees (at least 1.5 × tree heights apart to avoid the possibility of collecting seed from adjacent trees) with relatively abundant early-mature undehisced capsules (that may have been the product of flowering the previous year). Twelve trees satisfying the criteria were located (Table 1), though all had some older capsules from an earlier flowering. Samples of twigs containing the early-mature capsules were collected, allowed to dry and examined for the presence of viable seeds. Germination testing (using the method described below) of a sub-sample of all seed collected indicated that 81% of the seed was viable.

Supplementary data were obtained from two isolated E. albens trees — one at the Department of Environment and Climate Change’s research facility at Cowra (~380 m asl, MAR ~640 mm) Figure 1. A modified garbage bin seed collector (127/25) and another west of Molong (~620 m asl, MAR ~705 mm).

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 244 Seed fall and flowering in white box

Seed fall beneath the canopy was assessed at Walli by locating four collectors below the edge of the tree canopies at the four cardinal points of the compass. The total collection area beneath each tree represented 0.1–0.5% of the area beneath the canopy. At Cowra Total Total

(months) (months) and Molong, only two collectors, diagonally opposite, were located near the edge of the canopies. Recorded seed fall was therefore not from beneath the whole

To: canopy but rather from the ‘vicinity of the canopy edge’. Collectors were installed at Walli during August– September 1995 and shortly afterwards at Cowra and Molong.

Short-distance seed dispersal was assessed only at Walli from: by locating collectors 4 m beyond the mean diameter

Under canopy, canopy, Under of the canopy (i.e. 1 m inside the fence) at the four cardinal points of the compass at each of the 12 trees.

from: Monitoring seed fall

Outside canopy, canopy, Outside At Walli, seed fall was monitored from spring 1995 until autumn–winter 1997 for all 12 trees. Six of these

nd was alive in April 1999. in April alive nd was trees were monitored for a longer period: until February as alive in July 1997 but subsequently died. died. 1997 but subsequently in July alive as

995 but the latter was abandoned due to potential fire hazard. The groundstorey groundstorey The hazard. fire potential to due abandoned was latter 995 but the 1998 for one tree, and until April 1999 (a total of 43 months) for 5 trees. Seed fall at Cowra was monitored from September 1995 to April 1997 (19 months) and at Molong from October 1995 to April 1997 (18 months).

Collectors were emptied at approximately monthly

96 29 Oct 95 29 Aug 96 5 Dec 95 12 Sep 95 29 Aug 99 7 Apr 95 28 Sep 43 (initially 99 7 Apr 43 fortnightly) intervals by inverting the collectors e

e into a large funnel. The contents of each collector were 1996 groundstorey treatments treatments 1996 groundstorey period collection Seed – stored in a paper bag and later passed through a sieve with a mesh of 1.56 mm to extract coarser material. 1995

Herbicide Burn Herbicide Mow The finer material was examined under a binocular

microscope. Dark-coloured seeds were readily distinguishable from the smaller brown-coloured chaff. Seeds, including those with obvious insect damage, were counted and stored for germination testing, which Trunk Trunk 1.3 m (m) (m) 1.3 m

diameter at was carried out in a germination cabinet set at a regime of 12 h 30ºC day / 12 h 20ºC night. Covered Petri dishes, filled with wet cotton wool and overlaid with filter paper, provided the germination medium. Water was canopy canopy Approx. Approx. checked and topped up twice per week and germinants diameter (m) (m) diameter were counted weekly over a four-week period.

For each tree at each collection time, the amount of (m) (m) trees where seed fall and other monitoring was carried out between August 1995 and June 1999 June 1995 and August out between carried was monitoring other and fall seed where trees height height

Approx. Approx. seed was totalled and used to indicate the rate and time of seed fall. Seed numbers in each of the collectors ) ° outside the canopy were recorded separately and used to indicate the direction of seed fall. All seeds collected

Slope ( Slope (i.e. from under and outside the canopy) were tested ) Eucalyptus albens Eucalyptus for viability. ° 5 5 25 19 1.2 96 26 Mar 96 1 Apr 95 29 Aug 95 28 Sep 97 1 May 20 95 2 24 19 1.1 96 10 Oct 90 4 15 12 0.7 96 26 Mar 96 1 Apr 95 29 Aug 95 12 Sep 97 16 Jun 22 78 3 21 21 1.3 95 27 Sep 95 26 Oct 95 29 Aug 95 12 Sep 97 1 May 20 93 8 18 17 2.0 95 11 Sep 95 10 Oct 95 29 Aug 95 12 Sep 97 1 May 20 101 3 19 14 0.8 96 12 Nov 96 5 Dec 95 29 Aug 95 12 Sep 99 7 Apr 43 116 7 24 18 1.1 95 11 Sep 95 10 Oct 95 29 Aug 95 12 Sep 99 7 Apr 43 103 0 9 20 21 1.1 96 10 Oct 96 29 Oct 95 29 Aug 95 12 Sep 98 24 Feb 30 00 0 0 0 330 110 8 2 24 19 19 17 Apart 0.9 from 1.1 96 12 Nov occasions 96 12 Nov when 96 5 Dec collectors 95 29 Aug 95 28 Sep were 99 7 Apr damaged, 43 Aspect (compass (compass bearing the total seed collection area remained constant at Molong and Cowra but varied at Walli, where the Details of 14 of Details maximum seed collection area ranged from 11.2 m2 at

a b c d the start when all 12 trees were monitored to 4.7 m2 at 272 281 Molong Cowra n.a. n.a. n.a. n.a. 95 14 Oct 95 28 Sep 97 28 Apr 97 28 Apr 18 19 275 280 278 105 3 24 19 1.2 96 10 Oct 96 29 Oct 95 29 Aug 95 28 Sep 97 25 Aug 24 273 276 277 305 4 16 11 1.0 95 27 Sep 95 26 Oct 95 29 Aug 95 12 Sep 97 1 May 20 270 271 Tree no.Tree 274 279 Trees where seed fall monitoring was extended beyond mid-1997 are shown in bold. in bold. shown are mid-1997 beyond extended was monitoring fall seed where Trees One seedling was observed near the edge of the canopy in July 1996. It was protected from the second herbicide and mowing and w and mowing herbicide second the from protected was 1996. It in July canopy the of edge the near observed was seedling One One seedling was observed beneath the canopy in March 1996 but recorded as dead in late May 1996. May in late dead as 1996 but recorded in March canopy the beneath observed was seedling One One seedling was observed outside (SSE of) the canopy in November 1996. It was protected from subsequent herbicide and mowing a and mowing herbicide subsequent from protected was 1996. It in November canopy the of) (SSE outside observed was seedling One 1 in November groundstorey the of burning by followed be to intended 1995 was 24 October on herbicide of application earlier An 1.Table a b c d e was re-herbicided in November or October 1996, followed by mowing. mowing. by 1996, followed October or in November re-herbicided was the end when only 5 trees were monitored.

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 W.S. Semple, T.B. Koen and J. Henderson 245

Monitoring of non-seed material at Walli tion followed by mowing to disperse the thatch of collapsed grass and expose some bare ground. All exclosures were A wide range of non-seed material (e.g. mature and immature searched for seedlings at regular intervals, though this became capsules, flowers, floral buds, opercula, chaff and insect frass) progressively more difficult as exotic herbage rapidly was also found in the seed collectors. Opercula (as an indicator recolonised the burnt and/or herbicide-treated areas. of flowering from February 1996) were counted, and chaff (as an indicator of capsule dehiscence) was qualitatively assessed Data analysis on an abundance rating of 0 = nil to 3 = abundant. In some cases, only basal parts of reproductive structures were present. These Due to differing numbers of functional seed collectors at each were not counted as they could not be reliably attributed to buds, tree and varying time intervals between collections, total seed flowers or capsules. Where damaged structures could be reliably collected below the canopy at each tree on each occasion was identified, they were included in the counts of buds or flowers. converted to the number of seeds per square metre per week. As these aborted structures were considered peripheral to the The variable seed collection periods were represented by variable main study, only a sample of the data is presented. bar widths on histograms of seed fall. As a general indicator of the occurrence of capsule dehiscence over time, chaff abundance, In anticipation of the removal of some of the collectors, visual averaged across all collectors at each tree at each observation, monitoring of bud and flower abundance on all trees commenced was superimposed as a line graph on the seed fall histograms. in autumn 1997. Abundance was rated on a five-point scale: 0 = none, 1 = one to very few, 2 = scattered or a few small clumps, During the period when the two methods of assessing flower 3 = flowers obvious and dispersed across most of the canopy, abundance overlapped, correlations were not high, primarily due and 4 = abundant to maximum possible. All ratings were made to one measure (numbers of opercula in collectors) being an by the same observer. Final observations were made in early July accumulation over time and the other (visual abundance ratings) 1999, 3 months after the last of the collectors was removed and an instantaneous assessment. Both sets of data are presented using 41 months after flowering records commenced. As some trees an intuitive scale for opercula numbers per square metre per week had not commenced flowering when observations ceased, the relative to the 0–4 abundance rating. As relative flower abundance abundance of buds and the past temporal pattern of flowering over time was the main interest, the data are presented as line (which varied between trees) were used to estimate subsequent graphs. flowering. The two flower abundance assessment methods, which overlapped for varying periods at individual trees, yielded a Due to zero or very low total seed falls outside the canopy at 7 of record of flowering from mid-February 1996 to July 1999, and the 12 trees at Walli, only 5 could be used for an analysis of by extrapolation until December 1999, that is for four complete directional differences in seed fall beyond the canopy. Total seed flowering periods. fall for these trees was analysed using an analysis of variance for a randomised complete block design, with the four cardinal Seedbed treatments to encourage natural recruitment at compass points forming the treatments and assigning the trees as Walli blocks.

Seedbeds were prepared on three occasions at Walli by burning Results herbicide-treated groundstoreys beneath randomly selected pairs of trees (Table 1) during periods of low fire danger (a reasonable Total seed fall and viability over time at Walli condition imposed by the landholder). None of the fires was sufficiently intense to scorch even the bases of the canopies and Aggregate seed fall from all trees during the first 2 y was low this approach was abandoned. The groundstoreys of the remaining relative to that recorded during the following 1.5 y (Fig. 2). Seed six trees (Table 1) were subjected to a one-off herbicide applica- fall began to increase in late 1997 and peaked in the summers of

50 100

40 ) –2 30 50 20

10 Density of seedfall (no. seeds m Viability (%) 0 * 1995 1996 19971998 1999 Year

Figure 2. Total seed collected beneath all trees at Walli from 29 August 1995 to 7 April 1999. Varying periods of seed collection are indicated by varying bar widths. * = missing data. The proportion (%) of the seed that was viable at each time of collection (excluding an unreliable value for July 1996) is shown as a line graph.

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 246 Seed fall and flowering in white box

1997–1998 and 1998–1999. Over the 1995 1996 19971998 1999 full observation period, viability of all 80 Tree 270 seed collected averaged 76% and was 40 20 usually in the range of 60–100%. 10 * Lower-than-average viability was 3 Tree 271 evident in seed collected between July 2 (0%) and October (22%) 1996, a period 1 when little seed was collected. 80 ** 0 40 Tree 272 Seed fall beneath the edge of the 20 10 canopy * 3 Tree 273 Seed fall was highly variable between 2 trees (Fig. 3) such that presenting the 1 data as means would be misleading. * 80 0 During the first 20 months when all 12 Tree 274

) 40 1 trees were monitored at Walli, 7 trees – 20 shed virtually no seed or very little seed. 10 * week

Of the remainder, seed fall peaks of 2 80 – Tree 275 variable magnitude occurred from 40 20 spring 1995 to autumn 1996 and, to a 10 lesser extent, from spring 1996 to ** * 3 Tree 276 autumn 1997. Seed fall below the trees 2 at Molong and Cowra was quite dis- 1 continuous even during the minor peaks. * 80 ** 0 All trees monitored beyond autumn 40 Tree 277 20 1997 at Walli shed seed, more-or-less 10 continuously, from summer 1997–1998 * 3 to autumn 1999 when observations Tree 278 2 ceased. This included three of the ‘shy 1 seeders’ (no. 271, 272 and 278) from * * * * * the earlier period. Tree no. 274 shed 80 0 Tree 279 seed for most of the time. 40 20 Mean chaff rating (0 = nil to 3 abundant) 10 Chaff, an indicator of capsule de- * 3 hiscence, was found in the collectors Tree 280 2

much more frequently than seed. It was Rate of seedfall beneath the edge canopy (no. seeds m present, albeit sometimes in low amounts, 1 * * at all trees during much of the obser- 80 0 40 Tree 281 vation period (Fig. 3). As would be 20 expected, abundant chaff was associated 10 with the presence of seed in the ** 3 Cowra collectors. 2 1 Seed fall outside the canopy at 80 ** 0 Walli 40 Molong 20 Seed fall 4 m beyond the mean canopy 10 edge was considerably less than beneath *** the edge of the canopy. Across the five 1995 1996 19971998 1999 trees where appreciable seed falls were Year recorded, average numbers of seeds in collectors at each of the four cardinal Figure 3. Seed fall under the edge of the canopy of 12 Eucalyptus albens trees at Walli (trees no. compass directions were not signifi- 270–281), one tree at Cowra and one tree at Molong that were monitored for variable periods cantly different for the full period of between September–October 1995 and early April 1999. Seed fall expressed as numbers per observation. Analysis of only the latter square metre per week (aggregate of up to four collectors per tree), with bar widths indicating variable times over which seed accumulated in collectors. * = missing seed fall and/or chaff data. part of the observation period (January The line graph represents the mean (n = ≤ 8 collectors at each observation) chaff abundance 1998 to April 1999 when most of the rating (0–3). seed fell) indicated that differences were

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 W.S. Semple, T.B. Koen and J. Henderson 247

Table 2. Total seed fall per 0.116 m2 collector at 4 m beyond mean canopy diameter in four directions averaged across five Eucalyptus albens trees. Differences between the four directions were not significant (P > 0.05) during either period of record.

Recording period North East South West All records (Sep. 1995–April 1999) 13.2 24.0 20.6 22.6 Main seed fall period (Jan. 1998–April 1999) 09.0 16.2 16.4 15.8

60 Floral buds 50

) Opercula 1

– Flowers 40

week 30 2 – 20

(no. m 10

Rate of fall structures 0 1996 19971998 1999 Year

Figure 4. Quantities of buds and flowers (including part flowers) shed beneath and around the edge of the canopy of tree no. 274 at Walli from early February 1996 to early April 1999. To assist in gauging the size of the crop of flowers present each season, quantities of opercula are also shown.

also not significant (Table 2). However, because of high abundant flowers over a short period or less-abundant flowers variability between trees and the small amount of seed collected, over a longer period, but again there were trees that produced this result should be treated with some caution. few flowers. In contrast, all trees flowered (or were presumed to do so from the abundance of buds present at the last observation) Bud production during 1999 — either abundantly for a short period or less abundantly but over a longer period. Inflorescence buds were usually evident by early November, and pin buds shortly afterwards. Buds were ‘plump’ by about Across the 4 y, differences in patterns of flowering were evident February. Very few inflorescence buds or young pin buds were between individual trees: found in the collectors, but older buds were commonly aborted (a) four trees (no. 270, 271, 273, 278) produced moderate to before flowering commenced. Quantities beneath one of the abundant crops of flowers every second year regularly flowering trees at Walli are presented in Figure 4.

Flowering at Walli

A flower abundance rating of 1 or 2, or opercula numbers below ~10 m–2 week–1 (unless over an extended period) indicated minimal flowering by any particular tree during any one year. Even a nil record of flowering did not necessarily indicate that no flowers were produced, as scattered flowers may have been missed between observations or the few opercula shed may have missed the collectors. Though occasional flowers could be present at any time, flowering mainly occurred between autumn and late spring. Some trees flowered for much of this period, but most had peaks in winter or spring. Some whole flowers were aborted, but many of those collected had the disc and part of the hypanthium severed — presumably by parrots before being discarded (Fig. 5). Flower losses at one tree are presented in Figure 4.

Flowering was minimal in 1996 for all trees except tree no. 274 Figure 5. A flowering branchlet from tree no. 281 at Walli in July (Fig. 6). In 1997, all trees produced flowers, though minimally 1997. The disc and part of the hypanthium have been severed in trees no. 275, 276 and 279. In 1998, a few trees produced (presumably by parrots) from nearly half of the flowers (128/26).

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 248 Seed fall and flowering in white box

(b) three trees (no. 275, 276, 279) 1996 1997 1998 1999 60 produced moderate to abundant Tree 270 crops two years in a row 40 (c) three trees (no. 274, 277, 280) 20 produced for three or more years 0 4 Tree 271 in a row 3 2 (d) two trees (no. 272 and 281) were 1 0 intermediate between (a) and (c). 160 Tree 272 120 )

Averaged across all trees, flowers were 1 80 relatively abundant in 1997 and 1999, – 40 0 4

low in 1996 and intermediate in 1998 week

2 Tree 273

– 3 (Fig. 7a). Years of abundant flowers 2 were associated with above-average 1 winter and spring rainfall in the 60 0 Tree 274 previous year (Fig. 7b). 40 20 Natural recruitment at Walli 0 4 Tree 275 3 Only three naturally-recruited seed- 2 lings were seen during the period of 1 observation (Table 1). Two were first 60 0 Tree 276 observed in autumn–winter 1996 40 beneath the canopy of two trees (no. 20 271 and 279) where the groundstorey 0 4 had been treated the previous spring; Tree 277 3 but both were dead when observations 2 ceased. The effect of seedbed treat- 1 ment, whether by burning, herbicide 60 0 Tree 278 or both, was relatively short-lived, with 40 exotic groundcover returning to 20 Flower abundance (mean rating, 0 = nil to 4 abundant) around 100% within 6 months. The 0 4 only survivor was first observed in Tree 279 3 spring 1996 outside the canopy of tree 2 no. 28, the only tree with a pre- 1 60 0 dominantly native groundstorey. It Total rate of collection opercula beneath and outside canopy (no. m Tree 280 emerged before the groundstorey was 40 treated with herbicide. 20 0 4 Discussion Tree 281 3 2 1 Though the main focus of the work 0 reported here was on seed fall and to a 1996 1997 1998 1999 lesser extent flowering of E. albens, the results are discussed in terms of Year the seed production sequence, at least Figure 6. Flowering abundance and duration at 12 Eucalyptus albens trees at Walli from early as far as it is known. February 1996 to early December 1999. Two methods of assessment, with varying periods of overlap, are shown: (1) solid circles indicate numbers of opercula, expressed as numbers per Buds and flowers square metre per week (aggregate of up to eight collectors per tree); (2) open circles indicate visual abundance rating (0–4) until early July 1999 and open triangles indicate estimated abundance Inflorescence buds became visually ratings based on bud abundance and previously observed patterns of buds flowering at each tree. evident in November with bracts shed Note that the y-axis scale for tree no. 272 differs from all the others. shortly afterwards exposing ‘pin buds’ that matured to ‘plump buds’ by about February. Many floral buds were aborted, consistent with observations in other eucalypts (Florence (1985), namely April–November. Although there was variability 1996). Despite reports of flowering commencing in January– between trees, peak flowering generally occurred during winter February (Costermans 1983; Schrader 1987) or of occurring or spring but in some cases extended across much of this period. between March and May (Brooker and Kleinig 1990), Most of the trees produced at least a few flowers every year. observations at Walli were consistent with those of Clemson Many flowers were destroyed by parrots or otherwise aborted.

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 W.S. Semple, T.B. Koen and J. Henderson 249

(a) )

1 1996 1997 1998 1999 – 40 4

week 3 2 – 20 2

1

0 0 0 = nil to 4 abundant) no. opercula, m Flower abundance (mean Flower abundance (mean rating,

(b) 300 250 200 150 100 50

Quarterly rainfall (mm) 0 1995 1996 1997 1998 1999 Year Figure 7. (a) Flower abundance, as assessed by opercula numbers and/or visual rating, averaged across all 12 trees at Walli; (b) quarterly (seasonal) rainfall at Cowra during and immediately preceding the period of observations. The thickened line indicates long-term seasonal means.

The average pattern of abundant flowering was every second seed was held within capsules for about 3 y before being shed, year (Fig. 7a) but some trees flowered abundantly in two or more mainly during the warmer months, over a period of a year or so. consecutive years (Fig. 6). Whether each tree had a distinctive Where peak flowerings occur in consecutive years, seed from a pattern of flowering independent of environmental conditions or number of capsule crops can be present in the canopy and it may whether it was modified by (or entirely dependent on) prevailing be shed simultaneously. environmental conditions could not be determined from only 4 y of observation. From 1996 to 1999, however, average flowering Timing and abundance of seed fall varied between individual abundance was positively associated with above-average winter trees. This is not uncommon in forest eucalypts and has been and spring rainfall in the previous year (Fig. 7). attributed (Jacobs 1955) to varying dominance, age, health and aspect of individual trees within a stand. To what extent these (or As noted earlier, averaging across all trees can be misleading. other) factors affected seed fall was not investigated. Some trees For example, in a 30-y study of flowering synchrony of four box– produced very little seed, at least during the first 2 y, whereas ironbark species in Victoria, a graph of average flowering intensity two trees produced seed for much of the time (Fig. 3). As suggested considerable synchrony between the species; yet actual suggested above, this may have been due to varying proportions overlap of flowering periods was low due to variable flowering of capsules of different ages within the canopies of individual of the species between years (Keatley et al. 2004). It should also trees. This suggestion was supported by the variable patterns of be noted that as trees at Walli were deliberately and not randomly flowering, that is, abundance and frequency, subsequently selected, overall patterns of flowering (and seed fall) should not observed. Peak seed falls tended to occur in the warmer months, necessarily be considered representative of what may occur in a consistent with Lawrence et al.’s (1998) short-term observations natural stand of E. albens. of E. albens in the central western NSW.

Viability of seed extracted from a sample of early–mature capsules Seed maturation and seed fall present on all trees in mid-1995 was 81%, somewhat less than In the first 20 months (spring 1995 to autumn 1997), very little that reported (97%, Burrows 1995) from capsules of presumed seed was collected from all but two trees even though the presence similar age, suggesting the presence of some immature seed. of chaff in collectors indicated that some capsule dehiscence was Viability of seed within collectors ranged between about 60% occurring in all trees. However, for the five trees that were and 100% (mean = 76%) over time, though lower values were monitored until autumn 1999, seed fall was markedly greater in recorded in mid- to late 1996, a period when few seeds were the subsequent 20 months. It is likely that had all trees been collected. Some seed was obviously damaged by insects when monitored for the full period, a similar pattern would have collected (though whether prior to seed fall or whilst in the occurred. Seed fall during the first 20 months may have been collector was unknown), but in other cases no damage was evident mainly from older capsules, rather than from the ‘abundant early– and/or seed appeared to be immature. As most trees were mature capsules’ for which trees were selected in mid-1995. The relatively isolated (i.e. spaced at least, but usually considerably crop of early–mature capsules (containing viable seed) appeared more than, 1.5 times the height of adjacent trees apart), it was to have shed seed mainly during 1998. This suggested that viable possible that seed quality and quantity may have been adversely

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 250 Seed fall and flowering in white box affected by isolation as reported by Burrows (2000) for seed As only three naturally-recruited seedlings were seen during the collected from E. melliodora trees. course of the observations, it is difficult to relate their occurrence to seed fall. However, if a late 1995 emergence is assumed (see Seed dispersal below), then Curtis’ (1989) estimates of minimum seed quantities for recruitment, in the order of 100–200 seeds m–2 in the 6 months Asymmetrical distributions of established seedlings have been prior to emergence, may be too high. Prior seed fall, aggregated reported to the south of parent trees and have been attributed to from September to December (data for a full 6-month period protection (‘shading’) afforded by the parent against frost and/or were unavailable) at the three trees where recruitment occurred, desiccation by Curtis (1989) in northern NSW and by Lawrence was ~10 (no. 271), ~20 (no. 279) and ~70 (no. 281) seeds m–2. et al. (1998) in central western NSW and north-eastern Victoria. However, no recruitment was observed at tree no. 274 where a On NSW’s southern tablelands, Edgerton (1996) reported an seedbed had been prepared in spring 1995 and aggregate seed additional possibility, that is protection of frosted seedlings from fall during this period exceeded 100 seeds m–2. It would seem, early morning sunlight and subsequent death from ‘cold-induced therefore, that ‘instantaneous’ seed fall (not necessarily as high photoinhibition’ (but no evidence of an adverse effect of early as 100 seeds m–2 but at least sufficient to satiate ants) at the time morning sunlight on seedling survival was evident in pot trials or of suitable rainfall, rather than an aggregation over a 6-month in the distribution pattern of E. albens at high altitude on the period, may be a more useful predictor of the likelihood of central tablelands of NSW — Semple and Koen 2005). However, recruitment. in SA and Victoria (Venning 1988; Dalton 1993), clumpiness has been attributed to asymmetrical dispersal of seed by prevailing Other factors involved in seedling recruitment winds at the time of seed fall as reported in forest situations (e.g. Cremer 1966). Availability of sufficient seed at an appropriate time is only one of the factors necessary for successful recruitment. Having No evidence was found for asymmetrical distribution of seed, observed abundant eucalypt seedlings in Australian Capital albeit over relatively short distances, at Walli. Very little seed Territory woodlands only when rainfall was above average for at was collected at ~4 m from the edge of the canopy and there least three successive years, Jacobs (1955) considered rainfall to were insufficient data to examine whether asymmetry may have be the most important of these factors. However, subsequent work occurred in particular seasons, for example those that experienced on surface-sown or naturally deposited seed in modified rural high winds at the time of seed fall. Even allowing for the environments (summarised in Table 4 of Lawrence et al. 1998) possibility of diminished seed fall close to the canopy edge, as suggested that, for successful recruitment following seed fall in reported by Cremer (1966) for an isolated ~76-m high E. regnans spring or early summer, above-average rainfall in spring and tree, the numbers of seeds collected at 4 m from the canopy edge December followed up by at least 80 mm in either January, were surprisingly low. Clearly a more targeted approach, that is February or March were necessary. An examination of monthly seed collectors located over a wider area, would have provided a rainfall recorded at Cowra (summarised in Fig. 7b) indicated that better test of asymmetrical seed dispersal. Though asymmetrical rainfall during the period of observation was likely to have been seed dispersal may provide an acceptable explanation of many suitable for natural regeneration only in 1995–1996. cases of clumped seedlings elsewhere, the data presented here, albeit limited by the low numbers of seed collected, suggests The presence of a suitable seedbed satisfies a number of require- that the ‘protection hypothesis’ is an equally likely explanation ments for successful regeneration: seed access to mineral soil, a — at least in central western NSW. lodgment site for the seed, and at least for a time, absence of herbaceous competition. Seedbeds created by fires have been Seed fall and seedling recruitment effective in high-rainfall forests (e.g. Florence 1996) and mallee (e.g. Wellington and Noble 1985a,b) but have yielded variable The main reason for monitoring seed fall was to determine results in woodlands, especially those with non-native ground- whether or not seed supply was likely to be limiting successful storeys (Curtis 1990). Similarly variable results have followed recruitment of E. albens in highly modified environments. At herbicide treatments (e.g. Whalley and Curtis 1993; Semple and least two issues are involved here. First, studies elsewhere (e.g. Koen 1997, 2003), though herbicide followed by slashing was Wellington and Noble 1985b) indicate that all seed shed is likely associated with higher numbers of naturally-recruited seedlings to be removed by seed-harvesting ants unless seed supply exceeds than a herbicide-alone treatment on the tablelands not far from the rate of removal. Second, other studies (e.g. Lawrence et al. the Walli site (Lawrence et al. 1998). In contrast, native 1998; Semple and Koen 2003) suggest that seedlings arising from groundstoreys may require little or no treatment to be an effective autumn–winter germinations are unlikely to survive in modified seedbed (e.g. Venning 1985; Curtis 1990) and this may explain woodlands due to competition from annual exotics that germinate the presence of the sole surviving seedling at Walli. at the same time; and as noted by Prober (1996), cool-season annual exotics are particularly common in the southern part of Apart from occasional incursions, cattle were excluded from most the distribution of E. albens woodlands. Hence, in central western of the trees at Walli but wingless grasshoppers, abundant in the NSW at least, abundant falls of seed in spring and early summer summers of 1995–1996 and 1996–1997, were not. When in are more likely to result in successful recruitment than falls in plague numbers, these insects have been observed to be autumn and winter — particularly if some warm-season rainfall particularly damaging to young eucalypts established by direct can be expected. seeding (Dalton 1993) or from tubestock (Semple et al. 1995).

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 W.S. Semple, T.B. Koen and J. Henderson 251

Of the three naturally-recruited seedlings observed, all were at Environment and Climate Change). Chris Nadolny and two least six months old when first noted in mid- to late 1996 and anonymous referees provided constructive comment on the were probably survivors from an earlier and larger cohort that manuscript. Many thanks for all your help at the time and your probably emerged in late 1995 or early 1996 when rainfall patience since. satisfied Lawrence et al.’s (1998) criteria. Low seedling survival was probably due to competition from exotic herbage, browsing References by grasshoppers and possibly ‘competition’ from the parent tree in the case of seedlings beneath or near the canopy. Boland, D.J., Brooker, M.I.H., Turnbull, J.W. and Kleinig, D.A. (1980) Eucalyptus Seed. CSIRO, Australia. Brooker, M.I.H. and Kleinig, D.A. (1990) A Field Guide to the Conclusions Eucalypts. Vol. 1. South-Eastern Australia. Inkata Press, Melbourne, Victoria. Floral buds of E. albens were evident in November and flowering Burrows, G.E. (1995) Seed production in white box (Eucalyptus albens) occurred between autumn and late spring of the following year, in the South West Slopes region of New South Wales. Australian with peaks usually evident in winter or spring. Although most Forestry 58, 107–109. trees produced some flowers every year, abundant flowering of Burrows, G.E. (2000) Seed production in woodland and isolated trees individual trees ranged from 2 y in a row, 3 y in a row to every of Eucalyptus melliodora (yellow box, Myrtaceae) in the South second year. Collectively, however, at least one and usually more Western Slopes of New South Wales. Australian Journal of Botany 48, 681–685. trees flowered abundantly in each of the 4 y of observation (see Clemson, A. (1985) Honey and Pollen Flora. Inkata Press, Melbourne Fig. 6), giving rise to potentially multi-aged seed crops, not only and Department of Agriculture, Sydney, NSW. in individual trees, but had the trees been together, in a stand. Costermans, L. (1983) Native Trees and Shrubs of South-Eastern Australia. Weldon, Sydney, NSW. Seed fall was also highly variable such that (assuming short Cremer, K.W. (1966) Dissemination of seed from Eucalyptus regnans. longevity on the soil surface) it would have been limiting for Australian Forestry 30, 33–37. seedling recruitment at most trees for periods of a year or more. Cremer, K.W., Unwin, G.K. and Tracey, J.G. (1990) Natural Collectively, however, some viable seed was shed in the warmer regeneration. In: Cremer, K.W. (ed.) Trees for Rural Australia. months of all years of observation (see Fig. 2). It appeared that Inkata Press, Melbourne, Victoria, pp. 107–129. the abundant crop of early–mature capsules (with viable seed) Curtis, D. (1989) Eucalypt re-establishment on the Northern Tablelands of New South Wales. MSc thesis, University of New England, for which the trees were selected in mid-1995 did not shed seed Armidale, NSW. until 1998. Curtis, D. (1990) Natural regeneration of eucalypts in the New England region. In: Sowing the Seeds. Greening Australia, Deakin, ACT, Very little seed was recorded at 4 m beyond the edge of the canopy pp. 7–16. and the limited data suggested that it was dispersed equally in all Dalton, G. (1993) Direct Seeding of Trees and Shrubs: A Manual for directions. This suggests that factors other than seed dispersal Australian Conditions. Primary Industries SA, Adelaide, SA. (e.g. protection from desiccation) may explain asymmetrical Edgerton, J. (1996) Tree planting in cold climates: lessons from distributions of seedlings around parent trees. However, because fundamental research. Australian Journal of Soil and Water of limited data, asymmetrical distribution of seed cannot be ruled Conservation 9, 37–42. Florence, R.G. (1996) Ecology and Silviculture of Eucalypt Forests. as an explanation in some situations. CSIRO Publishing, Collingwood, Victoria. Harden, G.J. (ed.) (1990–93) Flora of New South Wales. Vols 1–4. Despite exclusion of stock and preparation of seedbeds beneath NSW University Press, Kensington, NSW. trees early in the observation period, seedling recruitment was Hillis, W.E. and Brown, A.G. (eds) (1984) Eucalypts for Wood minimal. Though the quantity of seed shed was low at this time, Production. CSIRO, Melbourne and Academic Press, Sydney. other factors, particularly competition from exotic herbage, were Jacobs, M.R. (1955) Growth Habits of the Eucalypts. Forestry and more likely explanations for recruitment failure. Timber Bureau, Commonwealth Government Printer, Canberra, ACT. Keatley, M.R., Hudson, I.L. and Fletcher, T.D. (2004) Long term Acknowledgements flowering synchrony of box–ironbark eucalypts. Australian Journal of Botany 52, 47–54. Grant Fisher, Bronwen Jones, Tom Semple and members of the Lawrence, J., Semple, W.S. and Koen, T.B. (1998) Experimental Regional Environmental Employment Program at Cowra attempts at encouraging eucalypt regeneration in non-native provided field assistance at Walli. Les Dean and Ifeanna Tooth pastures of northern Victoria and central western NSW. monitored seed collectors at Molong and Cowra respectively. Proceedings of the Linnean Society of NSW 119, 137–154. Insights into the natural recruitment of eucalypts were obtained Nowland, A. (1997) Sustainable Management Strategy for Travelling from landholders and staff of the Departments of Agriculture Stock Routes and Reserves in Central Western New South Wales. Rural Lands Protection Boards Association of NSW, Condobolin, and Conservation and Natural Resources during a study tour (by Coonabarabran, Coonamble, Forbes, Molong and Nyngan, 196 pp. W.S.) to Victoria in 1995. Part of the work reported above was Penfold, A.R. and Willis, J.L. (1961) The Eucalypts: Botany, carried out while one of us (W.S.) was a postgraduate student, Cultivation, Chemistry and Utilization. Leonard Hill, London and advised by Peter Clarke and David Curtis, at the University of Interscience Publishers, New York. New England. Funding for the study tour and fencing materials Prober, S. (1996) Conservation of the grassy white box woodlands: at the Thring Pastoral Company’s property at Walli was provided rangewide floristic variation and implications for reserve design. by the Farm Forestry Project of the Soil Conservation Service of Australian Journal of Botany 44, 57–77. NSW (part of which is now within the NSW Department of

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252 252 Seed fall and flowering in white box

Schabel, J. and Eldridge, D.J. (2001) A Comparison of Roadside and Venning, J. (1988) Growing Trees for Farms, Parks and Roadsides. Paddock Vegetation in the Box Woodlands of Eastern Australia. Lothian Publishing, Melbourne, Victoria. Occasional Paper No. 7, School of Geography, University of New Wellington, A.B. and Noble, I.R. (1985a) Post-fire recruitment and South Wales. mortality in a population of Eucalyptus incrassata in semi-arid Schrader, N. (ed.) (1987) The Flora and Fauna of the Parkes Shire. south-eastern Australia. Journal of Ecology 73, 645–656. Parkes Naturalist Group, Parkes, NSW. Wellington, A.B. and Noble, I.R. (1985b) Seed dynamics and factors Semple, W.S. and Koen, T.B. (1997) Effect of seedbed on emergence limiting recruitment of the mallee Eucalyptus incrassata in semi- and establishment from surface sown and direct drilled seed of arid south-eastern Australia. Journal of Ecology 73, 657–666. Eucalyptus spp. and Dodonaea viscosa. The Rangeland Journal Whalley, R.D.B. and Curtis, D.J. [1993] Natural regeneration of 19, 80–94. eucalypts on grazing land on the Northern Tablelands of NSW, Semple, W.S. and Koen, T.B. (2003) Effect of pasture type on Australia. In: Gaston, A., Kernick, M.D. and Le Houerou, H.N. regeneration of eucalypts in the woodland zone of south-eastern (eds) Proceedings of the IVth International Rangelands Congress. Australia. Cunninghamia 8, 76–84. Montpellier, France, April 1991. Association Francaise de Semple, W.S. and Koen, T.B. (2005) Altitude, frost and the distribution Pastoralisme, Montpellier, France, pp. 581–584. of white box (Eucalyptus albens Benth.) on the Central Tablelands Windsor, D.A. (2000) A review of factors affecting regeneration of and adjacent slopes of NSW. Proceedings of the Linnean Society box woodlands in the Central Tablelands of New South Wales. of NSW 126, 171–180. In: Hobbs, R.J. and Yates, C.J. (eds) Temperate Eucalypt Semple, W.S., Koen, T.B. and Marshall, C.J. (1995) The use of Woodlands in Australia: Biology, Conservation, Management and herbicides in restoring tree cover to Bathurst granite soils. Restoration. Surrey Beatty and Sons, Chipping Norton, NSW, Australian Journal of Soil and Water Conservation 8(4), 38–44. pp. 271–285. Spooner, P.G., Lunt, I.D. and Robinson, W. (2002) Is fencing enough? Yates, C.J., Hobbs, R.J. and Bell, R.W. (1994) Factors limiting the The short-term effects of stock exclusion in remnant grassy recruitment of Eucalyptus salmonophloia in remnant woodlands. woodlands in southern NSW. Ecological Management and I. Pattern of flowering, seed production and seed fall. Australian Restoration 3, 117–126. Journal of Botany 42, 531–542. Venning, J. (1985) Natural regeneration: a South Australian perspective. Trees and Natural Resources 27(1), 9–11.

Australian Forestry 2007 Vol. 70 No. 4 pp. 242–252